aaa

1
PWM
ac mains
~
~
M
3~
w*
u
s
current
limiter
v/f curve
gradient
limiter
t
g
u
s
*
arg( )
u
s
*
u
s
*
Sensorless Speed and Position Control of Induction Motor Drives
Joachim Holtz, Fellow, IEEE
Electrical Machines and Drives Group, University of Wuppertal
42097 Wuppertal Germany
Abstract Controlled induction motor drives without mechan-
ical speed sensors at the motor shaft have the attractions of low
cost and high reliability. To replace the sensor, the information
on the rotor speed is extracted from measured stator voltages and
currents at the motor terminals. Vector controlled drives require
estimating the magnitude and spatial orientation of the funda-
mental magnetic flux waves in the stator or in the rotor. Open
loop estimators or closed loop observers are used for this pur-
pose. They differ with respect to accuracy, robustness, and sensi-
tivity against model parameter variations. Dynamic performance
and steady-state speed accuracy in the low speed range can be
achieved by exploiting anisotropic effects of the machine.
1. INTRODUCTION
AC drives based on full digital control have reached the
status of a mature technology. The world market volume is
about 12,000 millions US$ with an annual growth rate of 15%.
Ongoing research has concentrated on the elimination of
the speed sensor at the machine shaft without deteriorating
the dynamic performance of the drive control system [1].
Speed estimation is an issue of particular interest with induc-
tion motor drives where the mechanical speed of the rotor is
generally different from the speed of the revolving magnetic
field. The advantages of speed sensorless induction motor
drives are reduced hardware complexity and lower cost, re-
duced size of the drive machine, elimination of the sensor
cable, better noise immunity, increased reliability and less
maintenance requirements. The operation in hostile environ-
ments mostly requires a motor without speed sensor.
A variety of different solutions for sensorless ac drives have
been proposed in the past few years. Their merits and limits
are reviewed based on a survey of the related literature.
2. CONSTANT VOLTS-PER-HERTZ CONTROL
To avoid the excitation at the eigenfrequencies of induc-
tion machines, a gradient limiter reduces the bandwidth of
the command signal as shown in Fig. 1. The v/f characteristic
ensures that the stator flux is maintained at its nominal value.
A preset minimum value of the stator voltage accounts for the
resistive stator voltage drop.
Since v/f -controlled drives operate purely as feedforward
systems, the mechanical speed may w differ from the refer-
ence speed w
s
*. A load current dependent slip compensation
scheme can be employed to reduce the speed error.
Constant volts-per-hertz control ensures robustness at the
expense of reduced dynamic performance, which is adequate
for applications like pump and fan drives, and for low-cost
drives. These contribute a substantial share of the market for
sensorless ac drives.
3. ROTOR FIELD ORIENTATION
3.1 Model reference adaptive system
The model reference approach (MRAS) makes use of the
redundancy of two machine models of different structures that
estimate the same state variable on the basis of different sets
of input variables [2]. The stator model in the upper portion
of Fig. 2 serves as a reference model, the rotor model in the
lower portion as the adjustable model. Its tuning signal w

is
obtained through a proportional-integral (PI) controller from
a scalar error signal e, computed from the phase displace-
ment between the two estimated flux vectors, y

r
S
and y

r
R
. As
the error signal e gets minimized by the PI controller, the tun-
ing signal w approaches the actual speed of the motor.
The stator model uses a delay element instead of an inte-
grator to eliminate an accumulation of the drift error. An band-
width limiter at the input of the rotor model ensures equiva-
lence of the two models.
u
s
i
s
stator model
e
w
y
r

rotor model
t
1
lm
t j
r
r
s

1
t
l s
s
1
kr
y
r

S
y
r

R
1
2
t
r
t
1
Fig. 2 Model reference adaptive system for speed estimation Fig. 1 Constant volts-per-hertz control
IECON Roanoke VA, 2003
aaa
l t
mains
w
s
u
s
u
s
*
d
i
s
d
field stator
coordinates
B
A
speed controller
w
r
w
k
1
k
2
=
=
1 k r y
rd 0 s r
y
rd 0 r h
k
1
k
2
kq
M
3~
i w sl i
d
*
r
s s s q

i w l i
q
r
s s s d
+
*
i
d
*
i
d
i
q
*
i
q
u
q
*
u
d
*
t
s
'
e
jd
e
jd
i -controller
q
i -controller
d
w
*
PWM
~
~
3.2 Feedforward control of stator voltages
Okuyama [3] derives the commanded stator voltages from
a steady-state machine model as the basic reference signals.
Their components in rotor field coordinates are marked by
the shaded frames in Fig. 3. It is thus the machine which,
through its model, lets the inverter duplicate the actual termi-
nal voltages. This process can be characterized as self-con-
trol.
A superimposed i
d
-controller compensates for the parame-
ter inaccuracy of the steady-state model, while an i
q
-control-
ler ensures the correct alignment of the rotor field angle with
the rotor flux vector. Transient deviations from the correct
field alignment are eliminated by the additional correction sig-
nals A and B [1, 3].
Torque rise time of this scheme is reported around 15 ms;
speed accuracy is within 1% above 3% rated speed and 12
rpm at 45 rpm [3].
3.3 Rotor field orientation with improved stator model
A sensorless rotor field orientation scheme based on the
stator model is described by Ohtani [4]. The upper portion of
Fig. 4 shows the classical structure in which the controllers
for speed and rotor flux generate the current reference vector
i
s
* in field coordinates. This signal is transformed to stator
coordinates and processed by a set of fast current controllers
CR PWM. A possible misalignment of the reference frame is
detected as the difference of the measured q-axis current from
its reference value i
q
*
. This error signal feeds a PI controller,
the output of which is the estimated mechanical speed w . It is
added to an estimated value w
r
of the rotor frequency, com-
puted from the reference values i
q
*
and y
r
*
. The integration
of w
s
produces the field angle d

.
The stator model is used to estimate the rotor flux vector
y
r
. The band-limited integration by means of a first-order delay
entails a severe loss of gain in the estimated flux component
y
r1
at low stator frequency, and a large field angle error. This
tendency is compensated by an added flux component y
r2
,
obtained from the reference value y

r
*
(S)
in stator coordinates.
The latter component dominates the resulting flux vector y

r
at
low frequency, replacing it by its reference value y
r
*
in a
smooth transition. This results in y

r
y
r
*
at low frequency,
which deactivates the rotor flux controller in effect. Howev-
er, the field angle d

as the argument of the rotor flux vector is

still under control through the speed controller and the i
q
-con-
troller, although the accuracy of d

reduces. Field orientation

is finally lost at very low stator frequency. Only the frequen-
cy of the stator currents is then controlled. The currents are
forced into the machine without reference to the rotor field.
This provides robustness and certain stability, although not
dynamic performance.
As the speed increases from a very low value, rotor flux
estimation becomes more accurate and closed loop rotor flux
control is resumed. The correct value of the field angle is re-
adjusted as the q-axis current, obtained from the y
r
-estimator
Fig. 5, now relates to the correct rotor flux vector. The i
q
-
controller then adjusts the estimated speed, and in consequence
also the field angle, for a realignment of the reference frame
with the rotor field.
At 18 rpm, speed accuracy is reported to be within 3 rpm.
Torque accuracy at 18 rpm is about 0.03 pu. at 0.1 pu. refer-
ence torque, improving significantly as the torque increases.
Minimum parameter sensitivity exists at
1
=
r
[4].
3.4 Adaptive Observers
The accuracy of the open loop estimation models described
Fig. 3 Feedforward of stator voltages, rotor flux orientation
Fig. 4 Sensorless speed control based on direct i
q
-estimation;
rotor field orientation
Fig. 5 Rotor flux estimator for the structure in Fig. 4
mains
e
jd
w
s
i
s
field stator
coordinates
speed controller
w
r
w
u
s
i
s
*
(S)
CR-PWM
estimator
u
s
yr
i
q
y
r

flux controller
N
y
r
*
d

y
r
*
i -controller
q
w
*
y
r
*
t
r
l
m
D
1
i
s
i
d
*
i
q
*
i
q
*
~
~
M
3~
N
i
q
d

u
s
i
s
y
r
*
y
r
*
(S)
y
r2

y
r1

u
ir
y
r

y
r

x
2 2
+y
t
1
e
jd
t
s
'
t
1
1
t
r
s
1
2
D
aaa
speed adaptation
i
s
u
s
stator rotor
y
r

w
error
compensator
w
w
DT
e
k
r
t
r
r
s
s
1
r
t j
r
t
r
t'
s

i
s
l
m
1
2
Di
s
Di
s
r
G ( w)
s
G ( w)
in the previous chapters reduces as the mechanical speed re-
duces. The limit of acceptable performance depends on how
precisely the model parameters can be matched to the corre-
sponding parameters in the actual machine. It is particularly
at lower speed that parameter errors have significant influ-
ence on the steady-state and dynamic performance of the drive
system.
The robustness against parameter mismatch and signal noise
can be improved by employing closed loop observers to esti-
mate the state variables, and the system parameters.
3.4.1 Full order nonlinear observer
A full order observer can be constructed from the machine
model, shown in the upper portion of Fig. 6. The model out-
puts the estimated values i

s
and y

r
of the stator current vector
and the rotor flux linkage vector, respectively.
Adding an error compensator to the model establishes the
observer. The error vector computed from the model current
and the measured machine current is i
s
= ii

s
i
s
. It is used to
generate correcting inputs to the electromagnetic subsystems
that represent the stator and the rotor in the machine model.
Kubota et al. [5] select the complex gain factors G
s
(w

) and
G
r
(w ) such that the two complex eigenvalues of the observer

1,2 obs
= k
.

1,2 mach
, where
1,2 mach
are the machine eigen-
values, and k > 1 is a real constant. The value of k > 1 scales
the observer by pole-placement to be dynamically faster than
the machine. Given the nonlinearity of the system, the result-
ing complex gains G
r
(w

) and G
r
(w

) in Fig. 21 must depend on

the estimated angular mechanical speed w , [5].
The rotor field angle is derived from the components of
the estimated rotor flux linkage vector y

r
.
The speed signal w is required to adapt the rotor structure
of the observer to the mechanical speed of the machine. The
signal is obtained through a PI-controller from the current error
i
s
. In fact, the term y

r
i
s|
|
z
represents the torque error T
e
caused by the current error i
s
. If a model torque error exists,
the estimated speed signal w

gets corrected by the PI control-

ler, thus adjusting the w -input to the rotor model. The phase
angle of y

r
, which is the estimated rotor field angle, then ap-
proximates the true field angle that exists in the machine. The
correct speed estimate is reached when the current error i
s
,
and hence the torque error T
e
reduce to zero.
The control scheme is reported to operate at a minimum
speed of 0.034 p.u. or 50 rpm [5].
4. STATOR FIELD ORIENTATION
Control at stator field orientation is preferred in combina-
tion with the stator model. This model directly estimates the
stator flux vector. Using the stator flux vector to define the
coordinate system is then a straightforward approach.
A fast current control system renders the stator current vec-
tor a forcing function, making the electromagnetic subsystem
of the machine behave like a complex first-order system. It is
only characterized by the dynamics of the rotor winding.
Other than rotor field orientation, stator field orientation
does not inherently provide decoupled control of torque and
flux. In the signal flow graph Fig. 7, the torque command ex-
erts an undesired influence on the stator flux. Xu et al [6] pro-
pose a decoupling arrangement, shown in the left of Fig. 7, to
eliminate the cross-coupling between the q-axis current and
the stator flux. The internal influence of i
q
is cancelled by the
external decoupling signal, provided that the estimated sig-
nals and parameters match the actual machine data.
The decoupling signal depends on the rotor frequency w
r
.
An estimated value w
r
is therefore required. It is obtained from
the estimator Fig. 8, which also supplies the remaining un-
known system variables. The stator flux linkage vector is es-
timated by the approximate stator model. The angular veloc-
ity of the revolving field is then determined from the stator
flux linkage vector. Although w

s
is computed from the esti-
mated value y

s
, its value is obtained at good accuracy. The
reason is that the uncertainties in y

s
are owed to minor offset
and drift components in measured currents and voltage sig-
nals. These disturbances exert little influence on the angular
velocity at which the space vector y

s
rotates.
The stator field angle d

is obtained as the integral of the

estimated stator frequency w
s
. The angular mechanical ve-
locity of the rotor is computed as w = w
s
w
r
, where w
r
is
derived from the stator current vector i
s
in stator field coordi-
nates and from y

s
using the condition for stator field orienta-
Fig. 6 Nonlinear observer;
the model of the induction motor is shown in the upper portion
Fig. 7 Machine control at stator flux orientation using a dynamic
feedforward decoupler
r w
w
s
w
T
e
T
L
machine
torque command
flux command
i
q
i
d
w
r
y
s
t'
r
t
r
'
t'
r
t
r
tm
t
r
'
l
m
aaa
t
1
N
y
s

y
s

2
w
y
s

w
s
d

i
s
(F)
N
r w
jw
sy
s
~
~
D
1
2
l s
s
r
s
x
2 2
+y x
l
s
i
d
i
q
t'
r
1
t
e
jd
t
r
i
s
u
s
1
d

r w
w
s
y
s

D
stator flux estimator
-estimator r w
y
s

d
dt
i
s
u
th
u
*
u
*
u
s
jm
e
tion [1].
The signal flow graph of a sensorless drive with stator field
orientation is shown in Fig. 9.
5. PERFORMANCE AT VERY LOW SPEED
5.1 Stator resistance
The important information on the field angle and the me-
chanical speed is conveyed by the induced voltage of the sta-
tor winding, independent of the respective method that is used
for sensorless control. The induced voltage u
i
= u
s
r
s
i
s
is
not directly accessible by measurement. It must be estimated,
either directly from the difference of the two voltage space
vector terms u
s
and r
s
i
s
, or indirectly when an observer is
employed.
In the upper speed range above a few Hz stator frequency,
the resistive voltage r
s
i
s
is small as compared with the stator
voltage u
s
of the machine, and the estimation of u
i
is per-
formed with good accuracy. Even the temperature-dependent
variations of the stator resistance are negligible at higher speed.
If operated at frequencies above the critical low speed range,
a sensorless ac drive performs as good as a vector controlled
drive with a shaft sensor; even passing through zero speed in
a quick transition is not a problem.
As the stator frequency reduces at lower speed, the stator
voltage reduces almost in direct proportion, while the resis-
tive voltage r
s
i
s
maintains its magnitude; it becomes the sig-
nificant term at low speed. It is particularly the stator resis-
tance r
s
that determines the estimation accuracy of the stator
flux vector. A correct initial value of the stator resistance is
easily identified by conducting a dc test during initialization
[7]. Considerable variations of the resistance take place when
the machine temperature changes at varying load. These need
to be tracked to maintain the system stable at low speed.
5.2 Nonlinearity of the PWM inverter
At low speed, also the voltage distortions introduced by the
nonlinear behavior of the PWM inverter become significant.
They are caused by the forward voltage of the power devices.
This voltage can be modeled by an average threshold voltage
u
th
, and an average differential resistance r
d
. The differential
resistance appears in series with the machine winding; its value
is therefore added to the stator resistance of the machine model.
Other than this, the influence of the threshold voltage is non-
linear which requires a specific inverter model.
Representing the device forward voltage, the polarity of the
threshold voltage u
th
is uniquely determined by the direction
of current flow. Differences in magnitude between the for-
ward voltage of an active device and a recovery diode are
minor and can be neglected. The threshold voltage is then con-
stant. The contributions of the three phase components per-
mit defining a threshold
voltage vector u
th
. This
vector has a constant
magnitude, while its
phase angle changes by
+60 or 60 whenever
one of the phase cur-
rents changes its sign.
The stator voltage vec-
tor u
s
= u* u
th
there-
fore follows a distorted
and discontinuous tra-
jectory, even when the
reference signal u* of
the pulsewidth modula-
tor describes a circular trajectory. Fig. 10 shows an example
at motoring operation. The fundamental amplitude of u
s
is
then less than its reference value u*; it is larger at regenera-
tion.
The voltage trajectory exhibits strong sixth harmonic com-
ponents. Since the threshold voltage does not vary with stator
frequency as the stator voltage does, the distortions are more
pronounced when the stator frequency, and hence also the sta-
tor voltages, are low. The latter may even exceed the com-
manded voltage in magnitude, which then makes correct flux
estimation and stable operation of the drive impossible.
The inverter voltage vector u
th
is modelled by the structure
in Fig. 11. Its phase angle depends on the polarities of the
three phase currents, of which only six possible combinations
exist. These control the sector indicator sec(i
s
), which locates
the current space vector in one particular 60-sector of the
e
jd
ac mains
~
~
M
3~
PWM
j
e
d
w*
y
s
*
w
y
s

u
s
i
s
(F)
i
s
(S)
u
s
u
s
*
(S)
i
s
*
(F)
current controllers
speed controller
flux controller
decoupling signal
model
field stator
coordinates
Fig. 9 Stator flux oriented control without speed sensor
Fig. 8 Estimator for stator flux, stator field angle, mechanical
speed and rotor frequency
Fig. 10 Voltage distortion due to
the nonlinearity of the inverter
aaa
i
s
sec( )
ac mains
i
s
i
s
u
th
u
*
u
s
inverter model
to stator flux estimator
u
s

inv
u
*
PWM
~
~
M
3~
r
d

comp
u
u
th
2
complex plane. The magnitude of u
th
is 2u
th
for formal rea-
sons [1]. The resistive voltage drop of the power devices r
d
i
s
depends on the differential resistance r
d
. The resulting com-
pensation signal u
comp
accounts for the linear and nonlinear
voltage drops in the inverter.
5.3 Stator flux estimation
Using a compensated inverter makes the controlling refer-
ence voltage vector u* an undistorted and accurate estimate
u
s
of the fundamental stator voltage vector. This enables an
accurate estimation of the stator flux linkage vector even at
very low stator frequency. A signal flow graph of a stator flux
estimator is shown in Fig. 12.
The upper portion of Fig. 12 shows that the stator flux vec-
tor is obtained by integrating the induced voltage u

i
= u

s

r
s
i
s
, [8]. Using a pure integrator avoids the estimation error
and bandwidth limitation associated with a low pass filter.
The method necessarily incorporates the identification of a
time-varying vector u
off
that represents offset and drift volt-
ages. These are unavoidable when analog circuits are involved,
as in stator current acquisition that provides the signal i
s
.
The offset voltage estimator is shown in the lower portion
of Fig. 12. It integrates the induced voltage u
i
in a second,
parallel flux estimation channel. The minimum and the max-
imum values per fundamental period T of the resulting signal
y

1
, occurring at the respective time instants t
min
and t
max
, are
subsequently added, separately for their respective d-axis and
the q-axis components, to form the components of the dc off-
set vector u
off T
. It is apparent that the offset of a sinusoidal
signal is zero if its minimum and maximum values sum up to
zero. The offset vector is lowpass-filtered and then fed back
to the input of the lower flux integrator in Fig. 12. Its input is
forced to zero in a steady state, which is proof that u

off
repre-
sents the dc component of u
i
.
The induced voltage is further contaminated by harmonic
components,. These are introduced by errors in the acquisi-
tion of the fundamental current component i
s1
, [1], by incom-
plete compensation of the dead time effect [9], and by param-
eter errors of the nonlinear inverter model. The harmonics of
u
i
are detected by the ac disturbance estimator shown in the
center of Fig. 12. The estimator simply computes the instan-
taneous difference in amplitude between the estimated stator
flux vector y

s
and its reference signal y
s
*
. Proportional con-
trol by a gain constant k
1
is used for fast cancellation of the
harmonic components.
The estimated field angle d

is obtained as the argument of

the stator flux vector y

s
, and the estimated speed as the deriv-
ative of the field angle d

.
5.4 Stator resistance adaptation
An important measure to improve the low-speed perfor-
mance is the accurate on-line adaptation of the stator resis-
tance, which is the most relevant parameter in sensorless con-
trol. Kubota et al [10] use the observer structure Fig. 6 to
determine the component Di
s
i
s
/ i
s
of the error vector Di
s
in
the direction of the stator current vector i
s
, which is propor-
tional to the deviation of the model parameter r
s
from the
actual stator resistance. The identification delay of this meth-
od is reported as 1.4 s.
A faster algorithm relies on the orthogonal relationship in
steady-state between the stator flux vector and the induced
voltage [8]. The vector diagram Fig. 13 illustrates the princi-
ple. The space vector signals are processed in a synchronous
reference xy-frame C, aligned with the stator current vector
i
s
. The relation r
s
= (u
sx
u
ix
)/i
x
is easily verified, where u
ix
=u
i
.
sin (g d).
To facilitate the correct estimation of r
s
, the vector u
i
of
the induced voltage, obtained as the derivative of the stator
flux vector y

s
, must not depend on the stator resistance r
s
.
The reason is that r
s
is the variable to estimate. The estimated
stator flux is therefore derived from the instantaneous reac-
tive power, defined as q = u
s
i
s|
|
z
, which notation describes
the z-component of the vector product of the stator voltage
and stator current vector.
The estimated stator resistance value is then used as an in-
put signal r
s
to the stator flux estimator Fig. 12. It adjusts its
parameter through a lowpass filter. The filter time constant
T
f
= w
sR
t
f
is about 100 ms. The identification delay of this
method is about 350 ms [8].
Fig. 11 Compensation of the inverter nonlinearity
e
jd
1
1
d

w
u
s
y
s

y
s

y
1

i
s
u
i

y
1min, max
u
off T
u
off

u
off

u
ac

r
s

y
s
*
x
t
1
1
j
e
d
r
s

ac disturbance estimator
S

y
d q ,
min
t ,
max
t
y
s
*
offset estimator
k
1
Fig. 12 Offet-compensated stator flux estimator
aaa
i
s
u
i

u
s
y
s

r
s
i
s

0
jm
u
sy
g d
g
d
e
C
x
y
S
1
i

,
0 16
t
4 8 12 20 s
0
1
1
i

i
q
0
i

i
q
i

w
0
0.02
5.5 Low speed performance achieved by improved models
The oscillogram Fig. 14 shows the response to load step
changes of rated magnitude while the speed is maintained con-
stant at 5 rpm. This corresponds to operating at a stator fre-
quency of 0.16 Hz (w
s
= 0.003) during the no-load intervals.
Fig. 15 demonstrates persistent operation at zero stator fre-
quency. A load step of 120% rated magnitude demonstrates
that correct field orientation was maintained.
6. SENSORLESS CONTROL THROUGH SIGNAL INJECTION
Signal injection methods exploit machine properties that
are not reproduced by the fundamental machine model de-
scribed in Sections 3 and 4. The injected signal excites the
machine at a much higher frequency than that of the funda-
mental field. The resulting high-frequency currents generate
flux linkages that close through the leakage paths in the stator
and the rotor, leaving the mutual flux linkage with the funda-
mental wave almost unaffected. The high-frequency effects
can be therefore considered superimposed to, and indepen-
dent of, the fundamental behavior of the machine. High-fre-
quency signal injection is used to detect anisotropic proper-
ties of the machine.
6.1 Anisotropies of an induction machine
A magnetic anisotropy can be caused by saturation of the
leakage paths through the fundamental field. The spatial ori-
entation of the anisotropy is then correlated with the field an-
gle d, which quantity can be identified by processing the re-
sponse of the machine to the injected signal. Other anisotro-
pic structures are the discrete rotor bars in a cage rotor. A
rotor may be also custom designed so as to exhibit periodic
variations within a fundamental pole pitch of local magnetic
or electrical characteristics, for example the dimensions of
the rotor slots. Detecting such anisotropy serves to identify
the rotor position angle, the changes of which define the shaft
speed.
The case of a saturation-induced anisotropy is considered
first. The fundamental field saturates the stator and rotor iron
in the region of higher flux density, there producing higher
magnetic resistivity of the local leakage paths. A transient ex-
citation by an injected voltage u
tr
changes the leakage fluxes
in the direction of the forcing voltage vector, u
tr
= dy
str
/dt.
The transient currents change accordingly. Given the higher
magnetic resistance in the saturated region, the local leakage
paths require more magnetizing current than the neighboring
zones of less fundamental flux density. The space vector i
tr
of
the transient current therefore inclines in space towards the
saturated region as seen from the leakage flux vector y
str
.
The leakage inductance is therefore not constant in a satu-
rated machine. It varies with the angular displacement between
the leakage flux vector y
str
and the fundamental flux vector.
The scalar leakage l
s
inductance in the stator equation u
tr
=
l
s
di
tr
/dt of an isotropic machine converts into a transient in-
ductance tensor l
s
when an anisotropy exists. The stator equa-
tion becomes u
tr
= l
s
di
tr
/dt. The tensor l
s
reflects only the
fundamental component of the spatial saturation function. It
describes the physical phenomenon that the derivative di
tr
/dt
of the transient stator current has a different direction in space
than the voltage vector u
tr
that originates the excitation.
There is generally more than one anisotropy present in an
induction motor. The existing anisotropies have different spa-
tial orientations, such as the actual angular position of the fun-
damental field, the position of the rotor bars within a rotor bar
pitch, and, if applicable, the angular position within a funda-
mental pole pair of a custom designed rotor.
Multiple anisotropies make the transient inductance tensor
more involved. Fig. 16 is a visualization of the inverse induc-
tance tensor l
s
1
that describes the direction of a transient
current change as a function of the exciting voltage: di
tr
/dt
= l
s
1
u
tr
. It displayed over one electrical revolution of an
Fig. 13 Vector diagram illustrating the estimation of the stator
resistance; S marks the stationary reference frame (a,b), and C
marks the current reference frame (x,y)
Fig. 15 Persistent operation at zero stator frequency, followed by
a load step of 120% rated magnitude
t
0
2p
0
1
d

1
y

sa
y

sb
0 5 s 1 2 3 4
y

sb
y

sa
0
1
d

i
q i
q
0.1
0
w

Fig. 14 Constant speed operation at 5 rpm (f

1
= w
s
/2p = 0.16
Hz, w
s
= 0.003), with load step changes of rated magnitude
aaa
unloaded induction machine.
The quasi-circles are caused
by the rotor bar anisotropy;
their number corresponds to
the number of rotor bars per
pole pitch. They displace
along another, low-frequen-
cy trajectory under the influ-
ence of the saturation aniso-
tropy. Their shapes vary,
changing from a vertical el-
lipse through almost a circle
to a horizontal ellipse. This
is owed to the changing mag-
netic saturation of the leak-
age paths as the fundamental
field rotates.
The response to an inject-
ed high-frequency signal nec-
essarily reflects all anisotro-
pies, field dependent and rotor position dependent. While in-
tending to extract information on one particular anisotropy,
the other anisotropies act as disturbances.
6.2 Signal injection
The injected signals may be periodic, creating either a high-
frequency revolving field, or an alternating field in a specific,
predetermined spatial direction. Such signals can be referred
to as carriers, being periodic at the carrier frequency with re-
spect to space, or time. The carrier signals, mostly created by
additional components of the stator voltages, get modulated
by the actual orientations in space of the machine anisotro-
pies. The carrier frequency components are subsequently ex-
tracted from the machine currents. They are demodulated and
processed to retrieve the desired information.
Instead of injecting a periodic carrier, the high-frequency
content of the switched waveforms in a PWM controlled drive
system can be exploited for the same purpose. The switching
of the inverter produces a perpetual excitation of the transient
leakage fields [11]. Their distribution in space is governed by
the anisotropies of the machine. Measuring and processing of
adequate voltage or current signals permits identifying their
spatial orientations.
6.3 Injection of a revolving carrier
A polyphase carrier rotating at frequency w
c
is generated
by the voltage space vector u
c
= u
c
exp(jw
c
t). This signal is
added to the controlling voltage of the pulsewidth modulator,
thus creating the transient excitation of the machine. The mod-
ulation by the machine anisotropies reflects in a space vector
i
c
of carrier frequency w
c
, forming part of the measured sta-
tor current vector i
s
. It is separated by a filter from the funda-
mental current i
s1
of lower frequency, and from the switching
harmonics of higher frequencies.
The separated space vector i
c
consists of a positive sequence
component i
p
, and a number of components i
nj
that rotate at
the angular velocity
c
+ 2
j
, i.e. in negative directions. Of
the latter components, one particular must be separated to
extract the angular orientation
j
t of the selected anisotropy.
The others act as disturbances.
Rotating at the frequency of the carrier signal, the current
vector i
c
follows in fact elliptic trajectories. This is owed to
the existence of its positive sequence and negative sequence
components. There are as many ellipses superimposed as an-
isotropies exist.
The axis ratio of a saturation ellipse is l
sq
/l
sd
, a close to
unity value that ranges between 0.9 and 0.96 [12, 13]. It is
therefore difficult to identify the angular inclination of the
ellipse and thus determine the angular orientation of the anisot-
ropy. The characterizing component i
n
is very small, being
superimposed by the larger positive sequence current vector
i
p
, and contaminated by the effect of other anisotropies and
disturbances. Finally, all these signals are buried under the
much larger fundamental current i
s1
, and under the switching
harmonics.
To give an example, the current amplitudes i
p
/i
1

R
and i
n
/
i
s R
from [12], referred to the rated fundamental current i
s R
are shown in Fig. 17. The values are measured from an induc-
tion machine at zero fundamental excitation, i
1
= 0, such as to
avoid saturation generating an additional anisotropy. Howev-
er, the rotor has an engineered anisotropy of l
sq
/l
sd
= 0.91.
There are three categories of negative sequence currents:
The current i
2
at frequency w
c
+ 2w is caused by the engi-
neered rotor anisotropy. Its harmonic spectrum spreads be-
tween w
c
and w
c
+ 2w
max
when the machine speed w
varies between 0 and w
max
, where w
max
= 2p
.
10 Hz is an
assumed maximum value in Fig. 17. This frequency com-
ponent carries the speed information; its magnitude i
2
=
0.022 i
s R
is extremely low.
The current i
slot
at frequency w
c
+ N/p w is caused by the
discrete rotor slots; it extends over the frequency range w
c
to w
c
+ N/p w
max
, where N is the number of rotor slots
and p is the number of pole pairs.
The current i
u
at frequency w
c
originates from winding
asymmetries, and from gain unbalances in the stator cur-
0
A/Vs
e
jm
l
s
1
210
210
210
210
Fig. 16 The inverse inductance
tensor l
s
1
, describing satura-
tion and rotor bar anisotropies
Fig. 17 Measured spectral current components from an unexcited
machine having two anisotropies, operated in a speed range w = 0
... w
max
= 2p
.
10 Hz (measurement data taken from [11])
w
1
0.008
0.005
i
i = 0.22%
2
i = 7.2% p
0
w
c
w
c
2w
max
+
w
i = 0
1
w
c
w
max
+
N
p

w
c
i = 0.069%
slot
i = 0.27%
u
aaa
rent acquisition circuits. This disturbance is in very close
spectral proximity to the speed-related component i
2
; both
converge to the same frequency at w = 0. Also, i
u
> i
2
in
this example.
If this machine was fully fluxed and loaded, another nega-
tive sequence current i
sat
would appear at frequency w
c
+
2w
s
. Also this component has an extremely low spectral dis-
tance 2(w
s
w) from the component i
2
, where w
s
w is the
slip frequency.
The distribution of the significant negative sequence spec-
tra in Fig. 17 indicates that it is extremely difficult to separate
these signals by filtering [14].
6.4 Speed and position estimation based on anisotropies
Degner and Lorenz [12] use a dynamic model of the me-
chanical subsystem of the drive motor as part of a closed phase-
locked loop (PLL) for spectral separation. A signal flow graph
of the speed and rotor position estimator is shown in Fig. 18.
The carrier generated space vector i
c
is transformed to a +w
c
-
reference frame in which i
p
shows as a complex constant. Its
contribution is nullified through the feedback action of an in-
tegrator. The remaining signal i
n
contains all negative sequence
components. It is transformed to the w
c
-reference frame. This
transformation shifts the frequency origin in Fig. 17 to w
c
;
the negative sequence components then appear as low-valued
positive sequence signals.
The unbalance disturbance at frequency zero is compensat-
ed by an estimated vector i

u
= i
u
exp(jj
u
), and the distur-
bance generated by rotor slotting by an estimated vector i

slot
.
What remains is the current vector i

2
= i

2
exp j(2w t + j
2
),
representing the rotor anisotropy as a second harmonic. This
signal carries the important information, since 2wt = 2J is
twice the rotor position angle; j
2
is a phase displacement in-
troduced by signal filtering.
The mechanical system model in the upper right of Fig. 18
receives an acceleration torque signal, formed as the diffe-
rence between the electromagnetic torque T
el
and the load
torque T

L
, both being represented by their estimated values.
The feedforward signal T

L
serves to improve the estimation
dynamics. It is obtained by a separate load
model. The estimated angular velocity w of
the rotor is the integral of the acceleration
torque, where
m
is the normalized mechan-
ical time constant. Integrating w yields the
estimated rotor position angle J

.
The estimated angle

controls two an-

isotropy models. The upper model in Fig. 18
forms part of the PLL. It computes the phase
angle component 2w

t = 2J

of the negative
sequence current vector i

2
, while its magni-
tude i

2
and phase displacement j
2
are intro-
duced as predetermined constant parameters.
As commanded by the computed phase an-
gle error e = i
2
i

2|
|
z
, the PID controller forc-
es the resulting space vector i

2
to align with
its reference vector i
2
, thus establishing J

J as desired. The anisotropy model thus
serves to impress on the estimated current vector i

2
the same
rotor position dependent variations that the real machine,
through its inherent anisotropy, forces on the negative se-
quence current component i
2
.
The rotor slot related current vector i

slot
is estimated by the
anisotropy model in the lower portion of Fig. 13 in a similar
fashion. The vector i

slot
is used to compensate the undesired
disturbance i
slot
that forms part of i
n
.
The saturation-induced anisotropy is not modelled in this
approach, which limits its application to unsaturated machines.
Another problem is the nonlinearity of the PWM inverter
which causes distortions of the machine currents. These gen-
erate additional negative sequence current components that
tend to fail the operation of the position estimator [15]. A
general difficulty of all revolving carrier injection methods is
the extreme low signal-to-noise ratio which is less than 10
3
in the example of Fig. 17. This calls for special efforts to en-
sure that the low-level signals are sufficiently reproduced when
doing the analog-to-digital conversion of the measured cur-
rents [16].
The nonlinearity of the PWM inverter, commonly known
as dead-time effect, produces distortions of the pulsewidth
modulation whenever one of the phase currents changes its
sign. With the high-frequency carrier signal superimposed to
the modulator input, the stator currents are forced to multiple
zero crossings when the fundamental phase currents are close
to zero. The effect causes severe current distortions that the
established methods for dead-time compensation cannot han-
dle.
Being time-discrete events, the current distortions are dif-
ficult to compensate in a frequency domain method. An off-
line identification method was proposed by Teske [15]. The
compensation is done here using sets of time-variable pro-
files over one electrical revolution, one profile for every op-
erating point in terms of load and excitation level, and sepa-
rately for the respective d- and the q-component.
Current publications on revolving carrier methods show that
numerous side effects require the signal processing structures
to get more and more involved, while the dependence on pa-
Fig. 18 Speed and rotor position estimator using a PLL to identify
the response to an injected revolving carrier
i
2
e
jx
e
jx
stator
coord.
+
i
s
1
2
r
e
PID
T
L

1
mechanical subsystem
w
e
jx
2
model of built-in saliency
slot effect model
Np
t
f
t
m
i
slot

e
jx
-coordinates w
c
w
c
t w
c

e i
u
jj
u
j
2

j
slot

-coordinates w
c
i
n
J

K, T , T
i d
aaa
i
s
*
d

u
cd
u
cq
= 0
u
s
*
mains
i
s
BPF
d

field angle estimator

u
c
coswt
i
c
( ) X
i
c
( ) S
Z
x
Z
y
D
N
p
4
LPF
i
s1
e
jd
e
jd
PWM
~
~
e
jx
rms
e
jx
rms
u
c
( ) X
t
1
M
3~
rameters or on specific off-line commissioning procedures
persists.
6.5 Injection of an alternating carrier
Revolving carriers scan the whole circumferential profile
of all anisotropies that exist in a machine. This serves to de-
termine the characteristics of only one particular anisotropy.
The subsequent processing steps fully identify its spatial ori-
entation without using a-priori knowledge. An alternative class
of methods relies on injecting not a rotating, but alternating
carrier in a specific, though time-variable spatial direction.
The direction is selected in an educated guess to achieve max-
imum sensitivity in locating the targeted anisotropy. It is an
advantage of these methods that already existing knowledge
is just updated by acquiring an incremental error per sam-
pling period.
6.6 Balance of quadrature impedances
The approach of Ha and Sul [17] aims at identifying a field
angle while the machine operates at low or zero speed. Ac-
cording to the vector diagram Fig. 19, an alternating high-
frequency ac carrier signal of amplitude u
c
is added to the d-
axis component of the control input of the pulsewidth modu-
lator. The voltage signal excites the machine in the direction
of the estimated d-axis. This direction may have an angular
displacement g = d

d from the true d-axis, the location of

which is approximately known from the identification in a
previous cycle.
The injected voltage adds an ac component i
c
to the funda-
mental stator currents of the machine, represented in Fig. 19
by the space vector i
s1
. Owing to the anisotropic machine im-
pedance, the high-frequency ac current i
c
develops at a spa-
tial displacement g with respect to the true field axis of the
machine.
When the machine is operated in saturated conditions, its
impedance Z
c
at carrier frequency w
c
is a function of the cir-
cumferential angle a in field coordinates. The impedance has
a maximum value Z
d
in the d-axis, and a minimum value Z
q
in the q-axis. The identification of the d-axis is based on the
assumption of a symmetric characteristic Z
c
(+a) = Z
c
(a).
An orthogonal xy-coordinate system is introduced in Fig. 18,
having its real axis displaced by /4 with respect to the esti-
mated d-axis. Its displacement with the true d-axis is then
(/4 g).
The identification procedure is illustrated in the signal flow
graph Fig. 20, showing the current control system and the gen-
eration of the ac carrier in its upper portion. The shaded frame
in the lower portion highlights the field angle estimator. Here,
the measured stator current i
s
is bandpass-filtered to separate
the ac carrier current i
c
. The current i
c
and also the excitation
signal u
c
cos w
c
t are transformed to xy-coordinates, and then
converted to complex vectors that have the respective rms am-
plitudes and conserve the phase angles. The complex high-
frequency impedance Z
c
is formed which is a function of the
transformation angle d

/4; seen from the field oriented co-

ordinate system in Fig. 19, the transformation angle is (/4
g). For reasons of symmetry, the real and imaginary compo-
nents Z
x
and Z
y
would be equal if accurate field alignment, g
= 0, existed. A nonzero error angle g makes Z
x
increase, and
Z
y
decrease. Hence an error signal e = Z
y
(g) Z
x
(g) can be
constructed which adjusts the estimated field angle d

to an
improved value by means of a PI controller. Fig. 20 shows
that this angle is used for coordinate transformation. In a con-
dition of accurate field alignment, d

d, from which g 0
follows.
It is documented in [17] that the difference between the
impedance values Z
d
and Z
q
is small when the machine is
fully saturated. The reduced error sensitivity then requires a
high amplitude of the injected signal.
6.7 Evaluation of elliptic current trajectories
The carrier injection methods described so far suffer from
certain drawbacks. We have the poor signal-to-noise ratio and
the parameter dependence of the revolving carrier methods,
and the low sensitivity of the quadrature impedance method.
Fig. 20 Signal flow graph of a field angle estimation scheme
based on impedance measurement in quadrature axes
F
S
d
i
s1
i
c
d
jq
j
y
x
0
g
4
+g
p
p
4
g
Fig. 19 Vector diagram showing the injected ac carrier i
c
in
different reference frames; F: field oriented frame, S: stationary
frame
aaa
Linke [18] uses an alternating ac carrier injected in the esti-
mated direction of the d-axis to extract a saturation anisotro-
py signal at high signal-to-noise ratio. An alternating carrier
voltage can be decomposed into two identical components that
rotate in opposed spatial directions. The associated current
trajectories assume elliptic shapes under the influence of ex-
isting anisotropies [1]. The spatial orientations of these el-
lipses deviate from the true field axis as a function of the er-
ror angle g = d d

between the true and the estimated field

axis.
The signal flow graph Fig. 21 illustrates the field angle es-
timation scheme. The bandpass filtered carrier-frequency cur-
rent i
c
is rotated by w
c
t + d

which makes its positive sequence

component i
p
a complex dc value. A lowpass filter suppress-
es all accompanying ac components like negative sequence
currents and switching harmonics. The imaginary part of i
p
is
sin 2g, which is proportional to the error angle g = d d

for
small error values. This signal is sampled at about 1 kHz. It
feeds an I-controller to create the estimated field angle d

in a
closed loop. In doing so, reference is made to the injected
carrier signal to build the transformation term w
c
t + d

.
As the acquired signal is a dc value in principle, the sam-
pling frequency can be chosen independently from the carrier
frequency. This ensures good and dynamically fast alignment
with the field axis without the need of choosing a high carrier
frequency. Also the dynamics of the speed and torque control
system is not impaired as the carrier signal does not appear in
the torque building q-current component. Therefore, the mea-
sured q-current need not be lowpass filtered, as is required
when a rotating carrier is used. According to Fig. 21, such
filter is only provided for the component i
d
in the excitation
axis.
The method yields a signal of high signal-to-noise ratio,
permitting operation at low carrier level. A 100-mA carrier
current was found sufficient for field angle estimation in a
1-kW drive system.
6.8 High-frequency excitation by PWM switching
The switching of a PWM inverter subjects the machine to
repetitive transient excitation. The resulting changes di
tr
/dt
of the transient machine currents are influenced by the an-
isotropies of the machine, characterized by the transient in-
ductance tensor l
s
. The transient current derivative is sepa-
rated from the fundamental component by taking the associ-
ated zero sequence voltage l
s
di
tr
/dt from a wye-connected
stator winding [19]. This voltage is proportional to the char-
acteristic component of the anisotropy signal in that particu-
lar phase axis in which a switching has occurred. Reconstruct-
ing the complete spatial orientation of an anisotropy requires
therefore the evaluation of a minimum of two switching events
in different phase axes. The switching instants must be exe-
cuted within a very short time interval, such that the angular
orientation of the anisotropy remains almost unchanged.
The acquired axis components of the anisotropy signal com-
pose a rotor position vector p(J
N
). The angle J
N
indicates
the angular position of the rotor within one rotor bar pitch.
The oscillogram Fig. 22 shows a full revolution of p(J
N
). This
emphasizes the high spatial resolution that this method pro-
vides. Also note-
worthy is the high
level of the ac-
quired signals,
which is around
35 V. A full me-
chanical revolu-
tion is performed
when J
N
/N incre-
ments by 2p
where N is the
number of rotor
bars.
The anisotropy
signals are ob-
tained by instanta-
neous sampling as
speed-independent values. The influence of the saturation-in-
duced anisotropy can be suppressed by spatial filters [20].
Rotor position acquisition is possible at sampling rates of sev-
eral kHz [13]. The spatial resolution and the signal-to-noise
ratio are very high. This permits implementing precise incre-
mental positioning systems for high dynamic performance.
However, the incremental position is lost at higher speed when
the frequency of the position signal becomes higher than twice
the sampling frequency.
Other than continuous carrier injection methods, which are
frequency domain methods, PWM excitation constitutes a se-
quence of non-periodic time-discrete events, and hence re-
quires time-domain methods for signal processing. The ab-
sence of spectral filters enables a faster dynamic response.
The oscillogram Fig. 23 shows a positioning cycle that re-
quires maximum dynamics at 120% rated torque. The high
magnetic saturation during the acceleration intervals tempo-
rarily reduces the amplitude of the position signals; the posi-
ac mains
i
s
*
i
s
u
s
*
w
c
t
u
c
LPF
BPF
ip
i
c
field angle estimator
d

i
q
i
d
d d

2( )
d

i
s
w
c
t +d

u
c
M
3~
PWM
~
~
e
jd
e
jd
e
jx
e
jx
x
1
t
1
e
jx
x
Fig. 21 Signal flow graph of a field angle estimation scheme
based on injecting an alternating carrier
t
40 V
e
jm
20
p
J
N
Fig. 22 Trajectory of the rotor position
vector p(J
N
) within one rotor bar pitch
aaa
0
p/2
t
0 0.2 0.4 0.6 1 s
0
4V
0
4V
100 ms
J
p

tion accuracy remains unaffected, as the relevant information

is contained in the phase angles.
7. SUMMARY AND PERFORMANCE COMPARISON
A large variety of sensorless controlled ac drive schemes
are used in industrial applications. Open loop control systems
maintain the stator voltage-to-frequency ratio at a predeter-
mined level to establish the desired machine flux. They are
particularly robust at very low and very high speed, but satis-
fy only low or moderate dynamic requirements. Small load
dependent speed deviations can be compensated incorporat-
ing a speed or rotor frequency estimator.
High-performance vector control schemes require a flux
vector estimator to identify the spatial location of the mag-
netic field. Field oriented control stabilizes the tendency of
induction motors to oscillate at transients, which enables fast
control of torque and speed. The robustness of a sensorless ac
drive can be improved by adequate control structures and by
parameter identification techniques. Depending on the respec-
tive method, sensorless control can be achieved over a base
speed range of 1:100 to 1:150 at very good dynamic perfor-
mance. It is a particular attraction of the fundamental model
that runs on a simple hardware platform. Also stable and per-
sistent operation at zero stator frequency can be achieved with
these models, provided that all drive system components are
precisely modelled and their parameters correctly adapted. Ac-
curate speed estimation in the very low speed region, howev-
er, is difficult since the fundamental model becomes unob-
servable.
Improved low speed performance can be achieved by ex-
ploiting the anisotropic properties of induction motors. The
spatial orientations of the anisotropies are related to the field
angle, and to the mechanical rotor position. These can be iden-
tified either by injecting high-frequency carrier signals into
the stator windings and process the response of the machine,
or by making use of the transients that a PWM inverter gener-
ates. These methods have recently emerged. They bear great
promise for the development of universally applicable sen-
sorless ac motor drives.
8. REFERENCES
1. J. Holtz, Sensorless Control of Induction Motors, Proceedings
of the IEEE, Vol. 90, No. 8, Aug. 2002, pp. 1358-1394.
2. C. Schauder, Adaptive Speed Identification for Vector Control
of Induction Motors without Rotational Transducers, IEEE Ind.
Appl. Soc. Ann. Meeting, San Diego Ca 1989, pp. 493-499.
3. T. Okuyama, N. Fujimoto, T. Matsui, and Y. Kubota, A High
Performance Speed Control Scheme for Induction Motor without
Speed and Voltage Sensors, IEEE Ind. Appl. Soc. Ann. Meeting,
Denver Co. 1986, pp. 106-111.
4. T. Ohtani, N. Takada, and K. Tanaka, Vector Control of Induc-
tion Motor without Shaft Encoder, IEEE Trans. Ind. Appl., Vol.
28, No. 1, Jan/Feb. 1992, pp. 157-165.
5. H. Kubota, K. Matsuse, and T. Nakano, DSP Based Speed Adap-
tive Flux Observer of Induction Motor, IEEE Trans. Ind. Appl.,
Vol. 29, No. 2, Apr/Mar. 1993, pp. 344-348.
6. X. Xu and D. W. Novotny, Implementation of Direct Stator Flux
Oriented Control on a Versatile DSP Based System, IEEE Trans.
Ind. Appl., Vol. 29, No. 2, Apr/Mar. 1991, pp. 694-700.
7. J. Holtz and A. Khambadkone, Vector Controlled Induction
Motor Drive with a Self-Commissioning Scheme, IEEE Trans.
Ind. Electr., 1991, pp. 322-327.
8. J. Holtz and J. Quan, Sensorless Vector Control of Induction
Motors at Very Low Speed using a Nonlinear Inverter Model and
Parameter Identification, IEEE Trans. Ind. Appl., Vol. 39, No. 4,
2003.
9. J. Holtz, Pulsewidth Modulation for Electronic Power Convert-
ers, Proceedings of the IEEE, Vol. 82, No. 8, Aug. 1994, pp.
1194-1214.
10. H. Kubota and K. Matsuse, Speed Sensorless Field Oriented
Control of Induction Motor with Rotor Resistance Adaptation,
IEEE Trans. Ind. Appl., Vol. 30, No. 5, Sept/Oct. 1994, pp. 1219-
1224.
11. J. Holtz, On the Spatial Propagation of Transient Magnetic Fields
in AC Machines, IEEE Trans. Ind. Appl., Vol. 32, No. 4, July/
Aug. 1996, pp. 927-937.
12. M. W. Degner and R. D. Lorenz, Using Multiple Saliencies for
the Estimation of Flux, Position and Velocity in AC Machines,
IEEE Trans. Ind. Appl., Vol. 34, No. 5, Sept/Oct. 1998, pp. 1097-
1104.
13. J. Holtz, Sensorless Position Control of Induction Motors an
Emerging Technology, IEEE Trans. Ind. Electr., Vol. 45, No. 6,
Nov/Dec. 1998, pp. 840-852.
14. N. Teske, G. M. Asher, M. Sumner, and K. J. Bradley, Suppres-
sion of Saturation Saliency Effects for the Sensorless Position
Control of Induction Motor Drives under loaded Conditions,
IEEE Trans. Ind. Electr., Vol. 47, No. 5, Sep/Oct. 2000, pp.
1142-1149.
15. N. Teske, G. M. Asher, K. J. Bradley, and M. Sumner, Analysis
and Suppression of Inverter Clamping Saliency in Sensorless
Position Controlled of Induction Motor Drives, IEEE Ind. Appl.
Soc. Ann. Meeting, Chicago, Sept. 30 - Oct. 4, 2001.
16. F. Briz, A. Diez, and M. W. Degner, Dynamic Operation of
Carrier-Signal-Injection-Based Sensorless Direct Field-Oriented
AC Drives, IEEE Trans. Ind. Appl., Vol. 36, No. 5, Sept/Oct.
2000, pp. 1360-1368.
17. J.-I. Ha and S.-K. Sul, Sensorless Field-Oriented Control of an
Induction Machine by High-Frequency Signal Injection, IEEE
Trans. Ind. Appl., Vol. 35, No. 1, Jan/Feb. 1999, pp. 45-51.
18. M. Linke, R. Kennel, and J. Holtz, Sensorless Speed and Posi-
tion Control of Permanent Magnet Synchronous Machines,
IECON, 28th Annual Conf. of the IEEE Industrial Electronics
Society, Sevilla/Spain, 2002.
19. J. Jiang, Sensorless Field Oriented Control of Induction Motors
at Zero Stator Frequency, Ph.-D. Dissertation (in German), Wup-
pertal, 1999.
20. J. Holtz and H. Pan, Elimination of Saturation Effects in Sensor-
less Position Controlled Induction Motors, IEEE Ind. Appl. Soc.
Ann. Meeting, Pittsburgh, Oct. 13-18, 2002.
Fig. 23 Sensorless position control showing motor shaft displace-
ment of 90 at 120% rated transient torque; traces from top: shaft
position angle J, position signals p

and p

; constant values
indicate zero speed.